Food Policy 67 (2017) 1–11 Contents lists available at ScienceDirect Food Policy journal homepage: www.elsevier.com/locate/foodpol Editorial Agriculture in Africa – Telling myths from facts: A synthesis q Luc Christiaensen Jobs Group, World Bank Group, USA a r t i c l e i n f o a b s t r a c t JEL classification: Stylized facts drive research agendas and policy debates. Yet robust stylized facts are hard to come by, O12 and when available, often outdated. The 12 papers in this Special Issue revisit conventional wisdom on O13 African agriculture and its farmers’ livelihoods using nationally representative surveys from the Living O17 Standards Measurement Study-Integrated Surveys on Agriculture Initiative in six African countries. At O18 O33 times they simply confirm our common understanding of the topic. But they also throw up a number O41 of surprises, redirecting policy debates while fine-tuning others. Overall, the project calls for more atten- tion to checking and updating our common wisdom. This requires nationally representative data, and suf- Keywords: ficient incentives among researchers and policymakers alike. Without well-grounded stylized facts, they Agriculture can easily be profoundly misguided. Africa Ó 2017 The World Bank. Published by Elsevier Ltd. This is an open access article under the CC BY IGO Measurement license (http://creativecommons.org/licenses/by/3.0/igo/). Technology Structural transformation Market 1. Introduction ground-truthed description of reality. With stylized facts often constituting the very starting point of research itself, this is odd Stylized facts drive research agendas and policy debates. They at best. provide a sense of importance, help frame the inquiry and are used As a result, academic debates and policies rely too often on out- to galvanize resources. So, the notion that 60–80 percent of work in dated or poor quality statistics, or just unrepresentative case study African agriculture is done by women has often been quoted to evidence. In Jerven’s words (2016, p. 343): ‘‘. . . the numerical basis motivate a greater gender focus in agricultural research and policy- on which we study African economies is poorer than we would like to making. Similarly, the observation that one third of the world’s think.” Sometimes numbers have even evolved into zombie statis- food is lost post-harvest, is used to rally the world around a food tics, numbers that live a life on their own, with their empirical waste reduction agenda (Chaboud and Daviron, 2017). basis undocumented and their origins unknown, though widely Yet, robust stylized facts, systematically obtained with reliable accepted as conventional wisdom, such as the notion that 70 per- methodologies and comparable data across countries, and settings cent of the world’s extreme poor are women.1 Devarajan (2013) within countries, are often hard to come by. Partly this is because calls for urgent action to remedy such ‘‘statistical tragedy”. After some concepts are simply difficult to measure. Not everything all, research, policies and investments can only be as good and effec- important can come packaged as a neat statistic. In the face of pres- tive as the data and evidence informing them (Beegle et al., 2016; sure to produce statistics irrespectively, wrong-headed numbers Jerven, 2016). may arise. It partly also reflects the lack of regular, representative The topic of quality data and measurement has recently started and reliable data to compile these facts. Finally, preoccupation to receive more attention, in the literature and in policy circles, with causal inference often leaves few incentives to produce a especially for macroeconomic statistics (Jerven, 2013), literacy (UNESCO, 2015) and poverty (Beegle et al., 2016). But the need to revisit common wisdom applies equally to agriculture, and in q The findings, interpretations, and conclusions expressed in this paper are particular, African agriculture (Carletto et al., 2015a). The world entirely the author’s. They do not necessarily represent the views of the Interna- in which African agriculture operates has been changing dramati- tional Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the 1 governments they represent. http://www.politifact.com/punditfact/article/2014/jul/03/meet-zombie-stat-just- E-mail address: lchristiaensen@worldbank.org wont-die/ http://dx.doi.org/10.1016/j.foodpol.2017.02.002 0306-9192/Ó 2017 The World Bank. Published by Elsevier Ltd. This is an open access article under the CC BY IGO license (http://creativecommons.org/licenses/by/3.0/igo/). 2 L. Christiaensen / Food Policy 67 (2017) 1–11 cally over the past two decades, following robust economic growth, 2. Myths, materials, and methods rapid urbanization, and climate change. But the information base on African agriculture has been limited for a long time, often even The LSMS-ISA Initiative5 supports national statistical offices in lacking reliable statistics on basic metrics such as the country’s the collection of at least four rounds of nationally representative agricultural yields. More generally, translating economic concepts household panel survey data in eight African countries during into numbers, such as the notions of productivity, seasonality, 2008–20. The papers in this study mainly draw on the first rounds commercialization, or a households’ net food marketing position collected during 2009–2012 in 6 of these countries (Ethiopia, (net buyer/seller), remains intrinsically challenging,2 often requir- Malawi, Niger, Nigeria, Tanzania and Uganda). They cover more than ing special data that are not standardly collected at scale, forcing 40 percent of the population in SSA and most of its agro-ecological analysts to rely on outdated or case study evidence or proxy mea- zones. While this does not make them representative for SSA as such, sures instead. together they provide a broad picture of the emerging new reality, The household survey panel data collected under the Living and also allow for elucidating differences across settings. In these Standards Measurement Study–Integrated Surveys on Agriculture countries, a total of 31,848 households were interviewed, with sam- (LSMS-ISA) Initiative provide a unique opportunity to take up this ple sizes per country varying between 2716 (Uganda) and 12,271 challenge. Over the period 2008–2020, nationally representative households (Malawi), of which, on average, 76% were rural. Burkina surveys are to be conducted in 8 African countries, representing Faso and Mali have joined the Initiative more recently. Their survey 45 percent of Sub-Saharan Africa’s (SSA) population. In these coun- findings are not included here. tries, four or more waves of detailed information are collected on The LSMS-ISA initiative also presents a number of notable inno- households’ economic activities, their income and well-being, with vations on the World Bank’s Living Standards Measurement Study special attention to agriculture. They also include a number of (LSMS) surveys, which for some time provided important informa- methodological innovations such as data gathering at the individ- tion on the lives of Africans, their income, their economic activities, ual and plot level, enabling more gender disaggregated analysis. and their wellbeing. Most importantly, it strengthens the coverage The data are made publicly available one year after their of household agricultural activities—the Integrated-Surveys-on-A collection.3 griculture part of LSMS-ISA. The surveys are based on household An international consortium of researchers under the Agricul- samples and designed from the perspective of the household, not ture in Africa – Telling Myths from Facts project led by the World the farm. As a result, medium and large scale farms are only spar- Bank, with complementary financing from the African Develop- sely covered in practice (Jayne et al., 2016), even though techni- ment Bank,4 exploited the first rounds of these surveys to revisit cally represented in the sample. Information is gathered at both common wisdom on African agriculture and its farmers’ livelihoods the household and the plot level, covering every aspect of farmers’ in the areas of agricultural technology, market engagement and life—from the plots they cultivate, the inputs they use, the crops structural transformation. Studies were each time framed around a they grow, the time they allocate per plot, the harvest that is cross-country investigation of conventional wisdom in these areas. achieved, the way they market it, the amount they lose post- A total of 12 broad stylized facts and sub-facts on African agriculture harvest, and so on. and rural livelihoods were thus reviewed, the results of which are Second, in addition to the integrated approach to data collection, brought together in this special issue. data gathering takes place at highly disaggregated levels, at the plot This synthesis summarizes the key findings, including through a level, but also at the individual level, such as for time allocation and number of easily accessible and replicable tables and figures, and plot management. This enables a more refined, gendered perspec- reflects on their implications. It seeks to facilitate the policy dia- tive on agriculture and rural livelihoods. Third, the surveys make logue and further research efforts including updating as new infor- wide use of ICT-tools. Tablets are used for data collection, improv- mation from LSMS-ISA or related surveys becomes available. The ing the quality of data (Caeyers et al., 2012); households are geo- findings at times confirm conventional knowledge, as one would referenced using Global Positioning System (GPS) devices, enabling hope, and put it on more solid empirical footing. More often they further integration with other data sources, and plot size is mea- fine-tune our understanding. But they also reveal some myths sured by GPS as opposed to self-reporting, improving accuracy of and raise new issues. Overall, the findings underscore the high aca- land based statistics (Carletto et al., 2015b). Finally, individuals demic and policy return from investing in regular, nationally rep- (not just households) are tracked across survey rounds, opening a resentative data collection and continuous examination of host of new research areas such as the study of migration. conventional wisdoms. These four innovative features of the data (integration, individ- The synthesis proceeds as follows. The next section expands on ualization, ICT use and intertemporal tracking) not only help obtain the underlying data base and the methodological approach taken. a more refined insight in African agriculture and its rural liveli- This precedes a synoptic overview of the 12 wisdoms revisited hoods, they also help scrutinize conventional views that have so and the core findings obtained when submitting them to the data. far lacked an adequate information base to do so, such as the gen- Section three expands on each of them, including their implica- der patterns in agricultural labor allocation or the application of tions for agricultural and rural development policies. Section four joint input packages in practice, i.e. at the plot level. The nationally concludes. representative scope of the data and the great degree of standard- ization across countries in questionnaire design and survey implementation further facilitate cross-country comparison as 2 See for example Rao (2007, Ch3) on measuring seasonality. well as comparisons across settings within countries. 3 The Initiative is financed by a grant from the Bill and Melinda Gates Foundation, Given the core objective of establishing solid statistics and dis- together with other contributors, and managed by the Development Economics Data Group of the World Bank Group. Several other data initiatives are also underway to tinguishing myths from facts, the studies have been primarily remedy the agricultural data situation, such as the Global Strategy to Improve descriptive in orientation, focusing on a careful definition and Agricultural and Rural Statistics, and the ensuing regional Action Plans. empirical operationalization of the concepts at hand. Regression 4 Other participating institutions included the Alliance for a Green Revolution in analysis is mainly used to complement the findings, to check Africa, Cornell University, the Food and Agriculture Organization, the London School of Economics, the Maastricht School of Management, the University of Pretoria, the University of Rome Tor Vergata, the University of Trento, and Yale University. For a 5 detailed description of the project and its collaborators, see http://www.worldbank. For a detailed description and access to the data and their documentation, see org/en/programs/africa-myths-and-facts. http://www.worldbank.org/lsms. L. Christiaensen / Food Policy 67 (2017) 1–11 3 robustness and generate hypotheses, not for causal inference. For Table 1 this reason, the panel nature of the data has not been exploited Conventional wisdom about African agriculture: true or false? much here, and the focus has been limited to the first rounds Paper The myth–what is the issue? Myth or fact of the data. Cross-country comparisons were systematically I. Production and technology adoption undertaken, though only when data comparability permitted it to 1 African farmers use very low levels of modern Not generally do so. inputs true The twelve topics examined were chosen following a review of 2 Population growth and market access determine Not generally intensification true key policy documents and expert consultations, and because of 3 Given its profitability, fertilizer use is too low Not true in their salient nature in ongoing academic and policy debates, in Nigeria addition to the feasibility of the new LSMS-ISA data to address 4 Women provide the bulk of labor in African False and advance these debates. The papers speak to the prevailing, agriculture overarching notions that (1) Africa’s agricultural technology is II. Market engagement backward; (2) that smallholder engagement with input, factor 5 Factor markets are largely incomplete in Africa True 6 Land markets play a minor role in African Increasingly and product markets remains limited and (3) that Africa and its cit- development false izens are behind in the structural transformation of their econo- 7 Modern inputs are not financed through formal True mies, occupations and incomes. Obviously, the papers address credit only a small subset of the stylized facts related to these and other 8 Agricultural commercialization enhances nutrition False topics on African rural development. Nonetheless, the facts revis- 9 Seasonality in African food markets is fading False ited have been driving a number of the contemporaneous debates III. Structural transformation on African agriculture and in initiating this endeavour, the project 10 Labor is much less productive in agriculture False 11 Incomes among African farmers are under- Largely false also seeks to catalyze a process of continuous fact-checking mov- diversified ing forward.6 12 Household non-farm enterprises only exist for Largely true Table 1 provides a schematic summary of whether the conven- survival tional views reviewed in these areas do indeed stand the test of the data, and to what degree. Are they myths in today’s African farm- ing context, or realities? The first four papers of this special issue confront these conven- But the answer to the ‘Myth versus fact?’ question is certainly tional wisdoms with the data. They begin with an update of the more complex than suggested in this table. Farming and farming extent of input use (Sheahan and Barrett, 2017). Binswanger- behavior are complex, and the concepts and statistics we use to Mkhize and Savastano (2017) then assess the current input appli- describe it, are unlikely to be as cooperative as the table suggests. cation rates within the macro-context of Africa’s current popula- Reality also varies—across farming systems, regions and countries, tion density and market access. This is followed by a case study and over time. The studies reflect this complexity, and explore the of the actual profitability of fertilizer use in Nigeria (Liverpool- nuances that any answer to the question ‘myth versus fact?’ has to Tasie et al. (2017)), drawing attention to a core micro-economic exude. principle driving input adoption, namely profitability. The fourth paper explores the potential for increasing crop output from clos- ing the gender productivity gap (Palacios-Lopez et al., 2017b). 3. A micro-economic update on African agriculture African farmers do in fact use modern inputs, even though not always efficiently. According to common wisdom, farmers in 3.1. Backward technology Africa hardly use modern inputs such as inorganic fertilizer and other agro-chemicals, or mechanization and water control. Using The prevailing view about African agriculture is that technology data from over 22,000 households and 62,000 agricultural plots is backward, and changing only slowly. Africa is decades from the six LSMS-ISA countries, Sheahan and Barrett (2017) revi- behind Asia from this perspective. Farmers are slow to respond sit this record, offering a number of ‘‘potentially surprising” facts. to modern methods of farming such as the use of modern They find that fertilizer and agro-chemical use is more widespread inputs and mechanization, land improvement and irrigation. than is often acknowledged. One third of the cultivating house- Official estimates put cereal yields in SSA still only at about 1.5 holds in the LSMS-ISA countries apply inorganic fertilizer and the ton/ha on average (2012–2014), about half those in South Asia average unconditional nutrient application rate is 26 kg/ha (corre- (3.1 ton/ha) and a quarter those in China (5.9 ton/ha) (World sponding to 57 kg of total fertilizer/ha). This is twice the SSA aver- Bank, 2017). age of 13 kg of nutrients/ha during the same period, even though This is a major concern. Two in five Africans still live in extreme still only one fifth of the OECD average.7 But rates vary considerably poverty (Beegle et al., 2016) and boosting agricultural productivity across countries (and also across regions within countries). Use is is key for poverty reduction (Christiaensen et al., 2011). It is also highest in Malawi, Ethiopia and Nigeria, where more than 40 percent somewhat surprising as the Boserup-Ruthenberg (BR) theory of of cultivating households apply inorganic fertilizer, but much lower agricultural intensification (Boserup, 1965; Ruthenberg, 1980) in Niger and Tanzania (17%) and Uganda, where inorganic fertilizer would have predicted higher uptake and the technologies are gen- use is virtually nonexistent (Fig. 1). erally believed to be profitable (Byerlee et al., 2007). Substantial One in six farmers also uses agro-chemicals, rising to one in gender gaps in agricultural productivity (O’Sullivan et al., 2014), three in Nigeria and Ethiopia—related to the more widespread combined with the notion that African women perform most of use of herbicides in these countries.8 These rates are substantially the work in agriculture further suggests that shifting attention to higher than the available numbers in the literature, and come female farmers to close this gap could be an important avenue to somewhat as a surprise. It prompted follow up work, also using boost agricultural output. 6 7 Ongoing research on other, related stylized facts, not included in this special The latest official numbers for SSA (2013) put the average 17.5 kg/ha (World issue, are for example the notion that the majority of rural households are net food Bank, 2017). 8 buyers (Palacios-Lopez et al., 2017a), the notion that African youth is exiting See Tamru et al. (2016) for a recent analysis of the patterns, drivers, and labor agriculture en masse (Maiga et al., 2017) and the notion that droughts are the main productivity implications of the rapid expansion of herbicide use in smallholder hazard in African livelihoods (Nikoloski et al., 2017). cereal agriculture in Ethiopia over the past decade. 4 L. Christiaensen / Food Policy 67 (2017) 1–11 % Share of cultivating households % Share of cultivating households 100 100 using inorganic fertilizer in main using agro-chemicals 90 90 77 season 80 80 70 70 60 56 60 50 50 41 33 35 40 31 40 30 30 13 16 17 17 20 8 11 20 10 3 10 3 0 0 Fig. 1. Modern input use in SSA is not uniformly low. Note: Agro-chemicals include pesticides, herbicides and fungicides. Source: Sheahan and Barrett (2017). the LSMS-ISA data, which suggests a strong correlation of pesticide the length of fallow and induce the higher use of organic manure use with increased value of the harvest, but also with increased health and fertilizers to offset declining soil fertility, as well as invest- expenditures and time lost from work due to sickness. This draws ments in irrigation and mechanization. Together these offset the attention to the health implications of increased agro-chemical use negative impact of population growth on farm sizes, maintaining in SSA and the need for effective regulatory policies as areas for future or increasing per capita food production and farm income. Alterna- attention (Sheahan et al., 2016). There are also signs of improved seed tively, if the intensification triggered by population growth and use, especially for maize, with 24 to 41 percent of maize growers market access is insufficient to raise food production and farmers’ reporting seed purchases, though these are likely lower bounds.9 incomes, beyond those of their parents, agricultural involution Input use appears also no longer confined to traditional cash crops, would be observed instead (Geertz, 1963). with input intensification (fertilizer, pesticides, improved varieties) The authors set out to explore the relevance of the BR frame- now equally (and at times even more) common for the maize staple. work in understanding contemporaneous modern input use in But it’s not all good news. With only 32 percent of cultivating SSA (acknowledging that longer term panel data would be needed households in the LSMS-ISA countries owning and 12 percent rent- for proper testing). They find only partial support. They establish ing some type of farm equipment (less than 1 percent own a tractor) that fallow areas have virtually disappeared (on average the rate and less than 5 percent using some form of water control (2 percent of fallow in the six countries is 1.2 percent (Fig. 3)), an important of the cultivated area), the incidence of mechanization and irriga- finding that has not been systematically documented so far. They tion remains quite small. Farmers also fail to benefit from the syn- also observe a response of modern input use (fertilizer, agro- ergetic use of inputs, using them mainly as substitutes instead of chemicals and improved seeds) to their newly developed exoge- complements.10 For example, only about a third of households in nous measure of agro-ecological potential, which is correlated with Ethiopia who apply at least one of three synergetic inputs (inorganic population density, and controlling for market accessibility (as fertilizer, improved seeds and irrigation) apply two or more. This measured through their newly developed measure of urban grav- reduces to only 15 percent when plots are considered (Fig. 2). It ity), but not with other measures of intensification, such as irriga- points to the lack of agronomic knowledge or perhaps the underesti- tion or crop intensity. Overall, they conclude that the existing use mated complexity of joint input application, which under certain cir- of organic and inorganic fertilizer is insufficient to maintain soil cumstances might make it rational to use them as substitutes rather fertility when fallow practices cease. And the weak response of than as complements. Perhaps the biggest message of the study crop intensity and irrigation is also not consistent with the BR though is that the country setting is the main factor behind farmer framework. In light of the promising outcomes suggested by the input use—the policy and market environments really do matter. BR framework, the process of intensification across these countries Agriculture intensification remains below what increased appears to have been too weak according to the authors. population pressure and market access would suggest. Sheahan The conclusion by Binswanger-Mkhize and Savastano is consis- and Barrett (2017) highlight that the use of modern inputs is no tent with the prevailing notion of underutilization of modern longer universally low, especially for inorganic fertilizer and maize, inputs in Sub Saharan Africa. But it remains unclear what an a key staple. But in their paper, Binswanger-Mkhize and Savastano acceptable level of intensification should be, given Africa’s current (2017) are less sanguine about the current state of affairs. In light state of population pressure and domestic market access. Intensifi- of Africa’s increased population pressure and market access, they cation is clearly starting to happen in some of the more densely argue that higher degrees of agricultural intensification should be populated landlocked countries and areas within these countries, observed. That’s at least what the longstanding Boserup- and has also been accompanied by a decline in poverty, as in Ethio- Ruthenberg farming systems theory of agricultural intensification pia. Globalization and Africa’s resource boom of the past two dec- would suggest. In this view, a virtuous circle of intensification ades have further enabled governments and farmers to meet food emerges, whereby population growth and market access reduce needs through an expansion of food imports and rural-urban migration (as in Nigeria), which may also have raised the levels at which population pressure really starts to bite and governments start to act upon it.11 So is the glass half full with respect to Africa’s 9 Comparing crop variety assessments by farmers with DNA fingerprints Ilukor et al. (2017) show that improved and hybrid variety cultivation is more widespread than commonly assumed by farmers in Uganda, even though the improved cultivated 11 varieties are no longer pure. Gollin et al. (2016) document how Africa’s commodity boom during the 2000s 10 It is commonly thought that modern inputs are seldom adopted in isolation since fueled economic growth in many resource rich countries as well as urbanization, in there are important complementarities between particular sets of inputs making it particular the emergence of consumption cities, characterized by higher shares of advantageous to use them together (Yilma and Berger, 2006; Nyangena and Juma, imports (including of food) and employment in non-tradable services, as opposed to 2014; Abay et al., 2016). tradable manufacturing or services. L. Christiaensen / Food Policy 67 (2017) 1–11 5 Fig. 2. Synergies from joint input use are essentially foregone. Note: The areas of the circles proportionally represent population size relative to the full sample of cultivating households. The percentages in the circles are conditional on using any one of the three included inputs (i.e. exclude the population that does not use any of the three inputs) and are not weighted. Source: Sheahan and Barrett (2017). 100 level regressions with household fixed effects augmented with sev- 90 eral time invariant plot characteristics. The findings are partly due Share of land in fallow (%) 80 to the high transport costs involved in procuring fertilizers from 70 the nearest distribution points (Table 2). Setting these so-called 60 ‘‘last-mile(s)” procurement costs to zero, as if the fertilizer were 50 directly available on the farm, would raise the number of plots 40 where fertilizer use pays (AVCR > 1) to 85–90 percent, and increase 30 the percentage of plots that could gain from adding fertilizer to over 20 10 7.8 5.0 70 percent. The importance of the ‘‘last mile(s)” for fertilizer acqui- 1.0 3.0 0.1 1.2 0 sition costs has also been raised by Minten et al. (2013) who report Malawi Niger Nigeria Tanzania Uganda Average that farmers in Ethiopia living about 10 km away from a distribution center faced transaction and transportation costs (per unit) that Fig. 3. Fallow land has virtually disappeared, except in Tanzania. Source: Binswanger-Mkhize and Savastano (2017). were as large as the costs to bring fertilizer over approximately a 1000 km distance from the international port to the input distribu- tion center. agricultural intensification, as the findings by Sheahan and Barrett But the limited profitability of fertilizer use in the Nigerian sam- (2017), using the same the data, would suggest, or is it half empty, ple is also due to poor marginal yield responses. At an average of as Binswanger-Mkhize and Savastano (2017) would hold? about 7.7 kg additional maize per additional kg nitrogen, these Returns to fertilizer use are not always favorable—at least in are well below the marginal physical products observed in other Nigeria: Another, more direct way to assess whether modern studies (ranging between 10 and 20 in Kenya) or when research inputs are underutilized, is to examine their profitability. The management protocols are followed (rising to 50 in Malawi) notion that fertilizer use is too low is predicated on the assumption (Marenya and Barrett, 2009; Matsumoto and Yamano, 2011; that it is profitable to use fertilizer at higher rates than currently Sheahan et al., 2013; Snapp et al., 2014). In this context, Marenya observed (Byerlee et al., 2007). There is, however, surprisingly lim- and Barrett (2009) also point to the importance of good quality ited empirical evidence to support this. Liverpool-Tasie et al. soils for inorganic fertilizer to be effective. The efficiency of inor- (2017) examine the profitability of fertilizer use in maize produc- ganic fertilizer is for example low on soil with low organic matter tion in Nigeria, where fertilizer use is already relatively high. Pro- content which is needed to prevent run-off and leaching and for duction theory suggests two criteria to assess profitability of input gradual nutrient release. Efficient absorption of nutrients is simi- use. The first (and weaker) criterion holds that fertilizer use is prof- larly impeded when soil is too acidic. Both are common problems itable when the overall net return is positive, i.e. as long as the of African soils (Barrett et al., 2017). value of the average kg of maize produced per kg of fertilizer (i.e. While the relatively high inorganic fertilizer application rates the average value product, AVP) is higher than the price per kg of observed in Nigeria are exceptional across the LSMS-ISA countries fertilizer; the average value cost ratio (AVCR) is greater or equal (Sheahan and Barrett, 2017), the findings by Liverpool-Tasie et al. than one. The second (and more widely used) criterion holds that (2017), which, in Nigeria, are also confirmed for rice (Liverpool- fertilizer use is profitable, when it is optimal or profit maximizing, Tasie, 2016), underscore the need to better understand the agro- i.e. as long as the value of the additional maize produced per kg of ecological and market conditions under which inorganic fertilizer fertilizer (i.e. the marginal value product, MVP) equals the price of use in particular, and other agricultural technologies in general, fertilizer; the marginal value cost ratio (MVCR) equals one. are profitable. Also, in the absence of adequate ex post coping Application of these criteria to maize producers in the cereal- mechanisms, still higher returns will be needed for fertilizer (and root crop farming system12 in Nigeria suggests that current applica- other modern) input use to be profitable or optimal (Dercon and tion rates yield a negative return for almost half the plots (AVCR < 1) Christiaensen, 2011). The results reported here have abstracted and that only about half the plots would gain from expanding fertil- from risk considerations. They also underscore the importance of izer use (MVCR > 1). AVP and MVP estimates are derived from plot integrated interventions (access to input use, extension, and markets). 12 Maize is one of the three most important cereals grown in Nigeria along with Women do not provide the bulk of labor in African agriculture. sorghum and millet. Maize plots in the cereal-root crop farming system represent There is also a gender dimension to low modern input use, with about 60 percent of the plots in the study sample. application rates typically lower among female headed households 6 L. Christiaensen / Food Policy 67 (2017) 1–11 Table 2 Fertilizer use and fertilizer use expansion do not pay on about half of the maize plots in the cereal-root cropping system in Nigeria. Year Full acquisition cost Fertilizer available in the village Maize plots (%) for which net benefit from fertilizer use is positive for a risk neutral farmer (AVCR  1) 2010 51 86 2012 56 88 Maize plots (%) for which expanding fertilizer use is profit increasing for a risk neutral farmer (MVCR  1) 2010 49 70 2012 53 86 Source: Liverpool-Tasie et al. (2017). Fig. 4. Women do not provide the bulk of labor in African agriculture. Note: ⁄ Population weighted. Source: Palacios-Lopez et al. (2017). and on female managed plots (Sheahan and Barrett, 2017). This explains an important share of the 20–25 percent gender gap in agricultural productivity (O’Sullivan et al., 2014). Combined with Most recently, it has among others given rise to the (re)- the widely accepted notion that women provide the bulk of the introduction of fertilizer programs (see Jayne et al., 2013 for an labor in agriculture in Africa, regularly quoted to be 60–80 percent, assessment). The lack of smallholder market participation is con- this has been taken to suggest that closing the gender productivity sidered to be holding back progress in the fight against malnutri- gap could go a long way in boosting Africa’s food supply. But the tion (von Braun and Kennedy, 1994), with 38 percent of African basis for this statistic on women’s labor share in Africa’s agricul- children under the age of five still suffering from growth retarda- ture is basically unknown and has been questioned (Doss et al., tion (Beegle et al., 2016). A related manifestation of market failures 2011). is the presence of food price seasonality, a widely acknowledged, Exploiting the plot level labor input records for each household but little systematically documented and increasingly neglected member across the six LSMS-ISA countries, Palacios-Lopez et al. phenomenon (Devereux et al., 2011). (2017b) find that women contribute just 40 percent of labor input The following five papers in the special issue query the prevail- to crop production. The numbers are slightly above 50 percent in ing notions of continuing factor market imperfection (Dillon and Malawi, Tanzania and Uganda, but substantially lower in Nigeria Barrett, 2017; Deininger et al., 2017; Adjognon et al., 2017), limited (37 percent), Ethiopia (29 percent) and Niger (24 percent) commercialization and its effect on nutrition (Carletto et al., 2017), (Fig. 4). The difference in the contribution in Nigeria between the and food price seasonality (Gilbert et al., 2017). northern (32 percent) and southern (51 percent) regions is illustra- Factor markets in general don’t function well. The conven- tive and tallies with expectations. It confirms heterogeneity across tional wisdom sees African agriculture trading in missing or imper- and even within countries. Controlling for the gender and knowl- fectly functioning factor markets. Dillon and Barrett (2017) edge profile of the respondents does not meaningfully change the conclude that this is largely true. At the heart of this finding is predicted female labor shares. Across the different countries, there the simple observation that the number of working age people in are also no systematic differences across crops or activities. the household should not affect the amount of labor used on the The authors conclude that the female labor share statistics in farm if factor markets function well. If the size of the household Africa’s agriculture do not, as such, support a (universally) dispro- does affect the amount of labor used on the farm, clearly factor portionate focus on female farmers to boost crop production. They markets (not only labor markets) are either absent or functioning further highlight the need to use consistent metrics when analyz- poorly. This test goes back to Benjamin (1992) and has been ing the benefits and costs of different interventions, as the gender applied in a number of settings (Udry, 1999; LaFave and Thomas, productivity gaps are in fact calculated based on differences in land 2014). The authors apply it systematically across five LSMS-ISA productivity among male and female managed plots, as opposed to countries. differences in returns to labor. The agricultural labor shares are in They find a significant link between labor input and household fact irrelevant for such calculations, irrespective of their size. That size, across all countries. The link is further robust to the gender of said, there may be many other important reasons for investing in the household head, the distance from markets and the agro- raising female labor productivity in agriculture, such as female climatic environment, suggesting that rural factor market failure empowerment and improving the nutritional outcomes of chil- is pervasive and structural. Yet, they also find that rural factor mar- dren. Establishing this requires further research for which nation- kets are not generally missing in an absolute sense. On average ally representative and gender disaggregated household survey across countries, 29.4 percent of agricultural households rent/bor- data on time use and intra-household control of income and pro- row land, 38.9 percent hire labor and 23.7 percent take out a loan ductive resources will be key. The new LSMS-ISA survey rounds (Table 3). Market existence thus appears less of a problem than make important steps in this direction, creating promising oppor- market function. Further work is needed to unpack the sources tunities for future research on gender and agriculture in Sub- (e.g. labor, land or financial markets) and causes of the underlying Saharan Africa. market failures to help target the necessary interventions. But land markets already perform a useful role: Deininger et al. (2017) explore in greater depth and more directly, the extent 3.2. Poor market functioning to which farmers are engaged in land markets, and the nature of that engagement. They confirm that farming households are A second recurring theme in the academic and policy debates already more actively engaged in land markets than commonly on African agriculture and rural development is the poor function- assumed, especially in land rental markets (Table 3). Land sales ing of input, factor and product markets (Barrett et al., 2017). Land, activity remains limited, though information was only available labor and credit markets are considered largely absent, even for Niger and Uganda. 20 years after the structural adjustment era of the 1980–1990s, Rental market access proves to have significant and beneficial impeding modernization and commercialization of agriculture. effects for the equalization of land endowments and farm L. Christiaensen / Food Policy 67 (2017) 1–11 7 Table 3 While generally incomplete, factor markets are not generally missing. % agricultural households Ethiopia Malawi Niger Tanzania Uganda Average Rent/borrow land 32.7 24.9 30.9 19.6 38.7 29.4 Hiring labor 30.2 40.1 48.7 30.1 45.2 38.9 Take loan/access credit 27.5 13.3 – 13.3 40.8 23.7 Source: Dillon and Barrett (2017). productivity. It permits land-poor but labor-rich households to to which we return below. Broader (nonagricultural specific) rural raise their resource base by renting in land. It facilitates other development investments and policies will benefit agricultural farmers to diversify their activity by renting out their land and tak- development through different channels. In fact, the most common ing up non-farm employment (without the risk of losing their land purpose of credit to a farming family in Africa is to finance the assets). These are profound gains in a process of structural change. start-up costs of non-farm enterprises (or to finance consumption). The effects are strongest in Malawi, Nigeria and Uganda. This may be partially to help finance input purchases and increase The authors suggest that institutional reforms (especially agricultural productivity, an area for further investigation. within the legal framework) are needed and effective in strength- Market participation is widespread, but the extent of agri- ening the role that land markets play in enhancing farmer welfare. cultural commercialization remains limited, without clear ben- They especially call for a more differentiated and empirically efits for nutritional outcomes. Taking the share of the gross value grounded view of the reality farmers face on land, which should of crop sales to the gross value of total agricultural production, i.e. be combined with a thorough understanding of the institutional the crop commercialization index, as their measure of market par- context. They make a number of suggestions on how household ticipation or agricultural commercialization, Carletto et al. (2017) questionnaires could be improved to achieve this. find that farmers sell on average around 20–25 percent of their Farmers rarely use credit when purchasing farm inputs: The crop output (a bit less in Malawi, slightly more in Uganda and Tan- role of credit in rural transformation is well understood, but do zania). Conditional on sales the rates amount to 20, 40 and 33 per- African farmers make use of credit when purchasing modern cent in Malawi, Uganda, and Tanzania respectively, indicating that inputs? Adjognon et al. (2017) show that credit use for fertilizer, while most farmers sell some crops in these three countries, the pesticide or seed purchases is extremely low, across credit type marketed shares remain limited. Conditional on planting and sell- (formal, informal, tied), crop (food or cash crop) and countries ing, 11, 37 and 30 percent of the value of food crops is commercial- (Fig. 5). They estimate that on average only 6 percent of farmers ized in Malawi, Tanzania and Uganda respectively (Fig. 6). use any form of credit to buy these inputs—at least in the four Unsurprisingly, commercialization rates rise with harvest size, countries they cover (Malawi, Nigeria, Tanzania and Uganda). Lar- but they are not confined to the traditional cash crops (which are ger farms are more likely to use credit, but interestingly, even fully commercialized). And even though they are less likely to sell, there, the use of informal credit is found to be rare. Modern inputs when they sell, female farmers tend to sell larger shares. are primarily financed through cash from nonfarm activities and Using household and individual panel data, the authors further crop sales instead. explore the relationship between agricultural commercialization, welfare and nutritional outcomes. Conventional wisdom suggests that the more farmers commercialize their operations through increased product-market orientation, the better off they can become. Greater market-orientation of agriculture would therefore be expected to raise incomes, improve consumption, and enhance nutritional outcomes in rural households (Braun and Kennedy, 1994). The authors find little evidence of this in the three countries studied, underscoring that many factors other than the degree of agricultural commercialization intervene in shaping nutritional outcomes, including other agriculture related factors. The articles in the 2015 Journal of Development Studies Special Issue Vol 51, Issue 8 guest-edited by Carletto et al. (2015c), explore some of Fig. 5. Virtually all purchases of modern inputs are financed from non-credit sources. Source: Liverpool-Tasie et al. (2017). While it is well documented that formal bank credit is seldom available to African farmers for input purchases, the working hypothesis is that farmers use tied credit with output and input traders and other sources of informal credit to finance the purchase of external inputs, while processors front inputs or cash for inputs in case of contract farming and cash cropping. The findings pre- sented here contradict this, pointing to the important role of off- farm income and crop sales instead. While this should not be taken as proof of credit constraint as such, it does highlight the impor- tance of the nonfarm sector for agricultural modernization and the intimate links between agriculture and off-farm employment, Fig. 6. Food crop commercialization remains limited. Source: Carletto et al. (2017). 8 L. Christiaensen / Food Policy 67 (2017) 1–11 Fig. 7. Excess seasonality in maize prices remains substantial and widespread.Note: seasonal price gaps, the percent mark-up of the peak over the trough month price, are calculated for each maize market in each country; stars indicate the median seasonal price gap across the markets examined in that country, the box borders indicate the 80th and 20th percentile and the endpoints of the vertical line the maximum and minimum. The SAFEX gap is the seasonal price gap for white maize observed in the South African Future Exchange market, which represents the international reference market. Source: Gilbert et al. (2017). them, such as the role of diversity in crop production and livestock (Kaminski et al., 2016) further shows that the estimated food price ownership. Livestock seems to emerge as particularly important seasonality also translates into seasonal variation in caloric intake and positively linked to nutrition, with crop production and diver- of about 10 percent among poor urban households and rural net sity of production positively associated in certain contexts. Bio- food sellers. Together the findings suggest that the current aca- fortification also pays off. Given the current emphasis on demic and policy neglect of price seasonality is premature. nutrition-sensitive agriculture, a better understanding of the trans- mission channels between crop choice, agricultural market 3.3. Faltering structural transformation engagement and nutritional outcomes continues to be a research priority. Following the end of the commodity supercycle and the col- There is substantial excess seasonality in food prices. Although lapse in commodity prices since 2012, attention has shifted to it is commonly accepted that seasonality permeates African liveli- structural transformation as the driver for growth and poverty hoods, surprisingly little attention is paid to it. Because of the sea- reduction in Africa (Barrett et al., 2017). Given generally perceived sonal nature of agricultural production, one area where seasonality lower labor productivity in agriculture and under-diversification of manifests itself, is in food prices. But there is also very little sys- rural incomes, much is vested in accelerating the transition out of tematic evidence on the extent of food price seasonality, and what agriculture (Gollin et al., 2014). The extent to which (rural) house- is available, is largely dated. Better integration of domestic food hold non-farm enterprises can be part of the solution remains markets today may explain part of the neglect. Gilbert et al. unclear. Their productivity is generally considered low, but little (2017) conclude that ‘‘while we all know about seasonality, it is systematic evidence is available. very unclear precisely what it is we know.” The last three papers in the special issue revisit the evidence on The authors show that seasonality in staple crop prices can be the agricultural labor productivity gap (McCullough, 2017), docu- substantial. They do so using trigonometric and sawtooth models ment the recent patterns of rural employment and income diversi- to overcome some of the systematic upward bias in seasonal gap fication (Davis et al., 2017), and explore the prevalence, patterns estimates from the common monthly dummy variable approach. and performance of rural nonfarm household enterprises (Nagler This arises especially in shorter samples of 10–15 years, a phe- and Naudé, 2017). Together they provide key micro stylized facts nomenon which has so far gone unnoticed.13 Maize prices in the to inform policy directions for the structural transformation of 193 markets from the seven African countries studied are on average rural Africa. 33 percent higher during the peak months than during the troughs The agricultural labor productivity gap is smaller than com- (Fig. 7). For rice the price gap is on average 16.5. These seasonal mark monly assumed and mainly due to underemployment, not intrin- ups are two and a half to three times larger than in the international sic lower productivity in agriculture. One common view, reference markets. Seasonality varies substantially across markets, especially popular among macro economists, is that labor is intrin- but in virtually none of them is it lower than in the reference sically far less productive in agriculture than elsewhere in the markets. Seasonality does not explain much of overall price volatility economy, and that a great deal is thus to be gained from transfer- over the year. ring labor out of agriculture, i.e. from accelerating the structural The findings confirm the existence of substantial excess season- transformation. This view finds support in the national accounts ality, for which there may be a host of reasons, including poor post- which show that, in Africa, value added per nonagricultural worker harvest handling, lack of storage facilities, lack of market integra- is six times larger than the value added per agricultural worker. In tion as well as lack of coping capacity (possibly because of financial developing countries, the ratio is 3.5. market failure) inducing sell-low, buy-back-high behavior Yet, this does not account for differences in human capital and (Stephens and Barrett, 2011). Follow-up analysis in Tanzania income diversification across sectors. Recent work by Gollin et al. (2014) shows that adjusting for those factors would bring the ratio 13 Unlike most of the studies reported here, the main data source for this study are down to 3.3 for Africa (and 2.2 for developing countries). But even not the LSMS-ISA household surveys, but rather price data obtained in key markets this may be misleading. In addition to neglecting differences in across the seven countries covered, five of which are LSMS-ISA countries. capital use across sectors, these macro numbers do not account L. Christiaensen / Food Policy 67 (2017) 1–11 9 Table 4 Underemployment explains most of the agricultural labor productivity gap. Nonagricultural/agricultural output Ethiopia Malawi Uganda Tanzania Average Per person productivity gap 2.25 4.76 4.48 4.20 3.92 Employment gaps 2.66 3.30 2.10 2.22 2.57 Per-hour productivity gaps 0.85 1.44 2.13 1.90 1.58 Source: McCullough (2017). very well for production for own consumption, which in percent on average across countries). But this also holds in other developing countries makes up a substantial share of agricultural regions (85 percent), with little change observed across countries output as shown by the agricultural commercialization figures over time or by GDP. above. Using output measures from micro data instead, reduces Rural households also derive about two thirds of their income the gap by another 20 percent in the ten countries for which from on-farm agriculture, compared with one third (on average) Gollin et al. (2014) had data. in other developing countries (Fig. 8). But when differences in The paper by McCullough (2017) in this special issue takes the the level of development are taken into account (as reflected in argument one step further still. Instead of comparing (micro based) GDP per capita), Africa is not on a different structural trajectory output estimates per worker, she also uses detailed micro data on from the other developing regions. There are nonetheless some hours worked in both primary and secondary occupations to mea- important differences. sure and compare labor productivity per hour worked. Doing so Rural households in Africa are less engaged in wage employ- shrinks the labor productivity gap to 1.6 on average, across the four ment, both on and off the farm. With the exception of Malawi, countries she studies, compared with 3.9 when using output per where it contributes 15 percent of income, the share of agricultural worker (Table 4).14 This is because workers in agriculture work wage income is only five percent on average. This suggests that the fewer hours (700 hours per worker per year on average in the four second round effects from a food price increase through the agri- countries she studies) than those outside agriculture (1850 hours). cultural wage labor market (Ivanic and Martin, 2008) are likely Not only is the agricultural labor productivity gap not as large as limited in SSA, unlike in India (Jacoby, 2016) or Bangladesh commonly portrayed, it follows mainly from underemployment (Ravallion, 1990) where about a third of rural households are and not from intrinsic lower productivity in agriculture. involved in agricultural wage labor, contributing 15–20 percent This shifts the policy focus from getting people out of agricul- of income (Davis et al., 2017, Table A2–3). Far fewer households ture per se to making better use of labor in agriculture. In agricul- are also involved in nonfarm wage employment, even after ture, work hours are constrained or rationed. This is possibly controlling for the level of development, resulting in a small share because of the seasonal nature of crop production, and relatedly, of nonfarm wage income in total income (8 percent compared with the seasonal availability of agricultural labor, another dimension 21 percent in the rest of the world). Most off-farm income in Africa of seasonality which deserves further investigation. In such a case, is derived from informal self-employment. smallholder labor productivity could be raised by making more use Of course there are differences across African countries, and of their labor off season, either on or off the farm. In environments within countries, partly driven by agro-ecological potential and with favorable temperature, water availability, and product market access, which are discussed in more detail by the authors. demand, this can be done on the farm through the promotion of Yet the overall employment and income patterns discerned here irrigation and horticulture enabling multiple crops a year, or correspond to the picture emerging from the macro-data. There through diversification into livestock products such as poultry, has been structural transformation away from agriculture in SSA eggs or dairy, which are less seasonal. over the past two decades—the agricultural share in GDP fell from Where demand for (seasonal) off-farm labor exists, the focus 23 percent in 1995 to 17 percent in 2013, and the share of agricul- may be on reducing barriers to mobility, as even small initial travel tural employment likewise fell by an estimated 10 percentage costs (compared with the potential gains) may compound the points (Barrett et al., 2017). Yet, it has been towards self- effects of inexperience, uncertainty and credit constraints to pre- employment in (non-tradable) services, not wage employment in vent subsistence farmers from accessing it, as illustrated by tradable manufacturing (Rodrik, 2016). This is partly because of Bryan et al. (2014) in Bangladesh.15 Or, in the absence of such labor Africa’s commodity boom, which fueled the emergence of demand, the role for off-farm employment generation nearby and consumption cities, characterized by higher shares of imports thus secondary town development, becomes important in accelerat- and employment in non-tradable services (Gollin et al., 2016). ing poverty reduction, as discussed by Ingelaere et al. (2017) in Given their prevalence, this raises the question of whether Tanzania. informal rural household enterprises can serve as pathways out African households are not unduly tied to agriculture: The of poverty. common view is that families in rural Africa rely more on agricul- Households in rural Africa diversify into non-farm activities ture compared with other parts of the developing world. Davis mainly for survival. While the dominant form of off-farm income, et al. (2017) revisit this using comparable employment and income there is little systematic evidence on the prevalence, patterns and aggregates from 41 national household surveys (14 from SSA of performance of Africa’s rural non-farm enterprises. Nagler and which 6 are LSMS-ISA) from 22 countries (9 from SSA). They con- Naudé (2017) are the first to exploit the enterprise modules of firm that agriculture remains the mainstay of rural livelihoods in the LSMS-ISA survey to characterize systematically the rural SSA. Virtually all rural households have an on-farm activity (92 household enterprise landscape in SSA. They find that non-farm self-employment activities in the African household are indeed mostly oriented around survival. The evidence lies in the nature 14 These gap estimates are plausibly still upper bounds. They do not correct for of these activities: most are small, unproductive, informal house- cross-sectoral differences in human and physical capital. 15 hold enterprises, operated from home, without any non- They show how an US$ 8.5 grant or credit incentive (conditional on seasonal migration and equivalent to the round trip fare), induced 22 percent of households to household employees and often operating only for a portion of send a seasonal migrant, increasing consumption at origin by 30–50 percent and re- the year. They are concentrated in easy-to-enter activities, such migration 1–3 years after the program by 8–10 percentage points. as sales and trade, rather than in activities that require higher 10 L. Christiaensen / Food Policy 67 (2017) 1–11 Fig. 8. Rural Africa derives a much larger share of income from agriculture, a similar share from non-farm self-employment, but less from wages (in and outside agriculture). Note: ROW = Rest of the World. Source: Davis et al. (2017). starting costs, such as transport services, or educational invest- ture’s large labor productivity gap and women’s disproportional ment, such as professional services. labor contribution to agriculture). Many others call for a revision But obviously not all are just there for survival, and labor pro- (the link between agricultural commercialization and nutrition, ductivity differs widely. Especially rural and female-headed enter- the assumption of input profitability) or fine-tuning (the missing- prises, those located further away from urban centers (Fig. 9), and ness of factor markets and the extent of input use, land market oper- businesses that operate intermittently display lower labor produc- ations and income diversification). They also call attention to tivity compared with urban and male-owned enterprises, or enter- neglected topics (seasonality) and open new lines of inquiry and pol- prises that operate throughout the year. Rural enterprises exit the icy attention (the role of agro-chemicals, the lack of agronomic market primarily because of a lack of profitability or finance, and knowledge, the reasons behind agricultural underemployment). due to idiosyncratic shocks. Nonetheless, the authors also show And some simply stand up to the data, as one would hope (the sur- that when the conditions are right, households can seize the vival orientation of most nonagricultural rural household enter- opportunities for enhancing family income. When households are prises, and the virtual absence of credit use for input acquisition). better educated and have access to credit, they engage in agribusi- Methodologically, the findings underscore the power of nation- ness and trade throughout the year—not just in survival mode. The ally representative data and cross-country standardization and policy challenge is to create a business climate to foster such activ- comparison. They also highlight the power of data innovation ities, which remains a tall order. and disaggregation at scale, as well as the power of descriptive statistics in shaping research and policy narratives. The hope is that the findings presented here catalyze further endeavors—to revisit our stylized facts in other areas and to update and deepen them as new data come along. Evidence-based policy-making requires sound facts as well as sound inference. With either one of them missing, researchers and policymakers alike risk flying blind. References Abay, K.A., Berhane, G., Taffesse, A.S., Koru, B., Abay, K., 2016. Understanding Farmers’ Technology Adoption Decisions: Input Complementarity and Heterogeneity. Ethiopia Strategy Support Program Working Paper 82, Ethiopian Development Research Institute and International Food Policy Research Institute: Addis Ababa, Ethiopia. Adjognon, S.G., Liverpool-Tasie, L., Reardon, T., 2017. Agricultural input credit in Sub-Saharan Africa: telling myth from facts. Food Policy 67, 93–105. Barrett, C.B., Christiaensen, L., Sheahan, M., Shimeles, A., 2017. On the Structural Fig. 9. Enterprise productivity (Uganda) declines with distance from the urban Transformation of Rural Africa. World Bank Policy Research Working Paper center. Source: Nagler and Naudé (2017). 7938, Washington D.C. Beegle, K., Christiaensen, L., Gaddis, I., Dabalen, A., 2016. Poverty in a Rising Africa. World Bank Group, Washington DC. 4. Concluding remarks Benjamin, D., 1992. Household composition, labor markets, and labor demand: testing for separation in agricultural household models. Econometrica 60–2, 287–322. Stylized facts drive much of our research and policies, but Binswanger-Mkhize, H.P., Savastano, S., 2017. Agricultural intensification: the robust stylized facts remain hard to come by, because of concep- status in six African countries. Food Policy 67, 26–40. Boserup, E., 1965. The Conditions of Agricultural Growth. Aldine, Chicago. tual challenges, lack of data or insufficient incentives. All three Braun, von J.V., Kennedy, E., 1994. Agricultural Commercialization, Economic apply when it comes to statistics on African agriculture and its Development and Nutrition. Johns Hopkins University Press. farmers’ livelihoods. The papers in this special issue have exploited Bryan, G., Chowdhury, S., Mobarak, A.M., 2014. Underinvestment in a Profitable the newly available nationally representative LSMS-ISA survey Technology: The Case of Seasonal Migration in Bangladesh. Econometrica 82–5, 1671–1748. data to begin addressing this void. They confront some of the more Byerlee, D., Kelly, V., Kopicki, R., Morris, M., 2007. Fertilizer Use in African salient conventional wisdoms on agricultural technology, markets Agriculture: Lessons Learned and Good Practice Guidelines, Directions in and farmers’ livelihoods with these data. Development. World Bank, Washington D.C. Caeyers, B., Chalmers, N., De Weerdt, J., 2012. Improving consumption The findings reveal how a few of our stylized facts on African agri- measurement and other survey data through CAPI: evidence from a culture, and the policies they motivate, are wrong headed (agricul- Randomized Experiment. J. Dev. Econ. 98–1, 19–33. L. Christiaensen / Food Policy 67 (2017) 1–11 11 Carletto, C., Joliffe, D., Banerjee, R., 2015a. From tragedy to renaissance: improving Liverpool-Tasie, L.S.O. et al., 2017. Is increasing inorganic fertilizer use for maize agricultural data for better policies. J. Dev. Stud. 51–2, 133–148. production in SSA a profitable proposition? Evidence from Nigeria. Food Policy Carletto, C., Gourlay, S., Winters, P., 2015b. From guesstimates to GPStimates: land 67, 41–51. area measurement and implications for agricultural analysis. J. Afr. Econ. 24–5, Maiga, E., Christiaensen, L., Palacios-Lopez, A., 2017. Are African Youth Exiting 593–628. Agriculture en Masse? World Bank: mimeographed. Carletto, C., Ruel, M., Winters, P., Zezza, A., 2015c. Farm-level pathways to improved Matsumoto, T., Yamano, T., 2011. Optimal fertilizer use on maize production in east nutritional status: introduction to the special issue. J. Dev. Stud. 51-8, 945–957. Africa. In: Emerging Development of Agriculture in East Africa. Springer, Carletto, C., Corral, P., Guelfi, A., 2017. Agricultural commercialization and nutrition Netherlands, pp. 117–132. revisited: empirical evidence from three African countries. Food Policy 67, 106– Marenya, P.P., Barrett, C.B., 2009. Soil quality and fertilizer use rates among 118. smallholder farmers in Western Kenya. Agric. Econ. 40–5, 561–572. Chaboud, G., Daviron, B., 2017. Food losses and waste: navigating the McCullough, E.B., 2017. Labor productivity and employment gaps in Sub-Saharan inconsistencies. Global Food Security 12 (March), 1–7. Africa. Food Policy 67, 133–152. Christiaensen, L., Demery, L., Kuhl, J., 2011. The (evolving) role of agriculture in Minten, B., Koru, B., Stifel, D., 2013. The last mile(s) in modern input distribution: poverty reduction-an empirical perspective. J. Dev. Econ. 96–2, 239–254. pricing, profitability, and adoption. Agric. Econ. 44–6, 629–646. Davis, B., Di Giuseppe, S., Zezza, A., 2017. Are African Households (Not) leaving Nagler, P., Naudé, W., 2017. Non-farm entrepreneurship in rural Sub-Saharan agriculture? Patterns of households’ income sources in rural Sub-Saharan Africa: new empirical evidence. Food Policy 67, 175–191. Africa. Food Policy 67, 153–174. Nikoloski, Z., Christiaensen, L., Hill, R., 2017. Droughts dominate Africa’s Risk Deininger, K., Savastano, S., Xia, F., 2017. Smallholders’ land access in Sub-Saharan Environment, World Bank: mimeographed. Africa: a new landscape? Food Policy 67, 78–92. Nyangena, W., Juma, O.M., 2014. Impact of Improved Farm Technologies on Yields: Dercon, S., Christiaensen, L., 2011. Consumption risk, technology adoption and The Case of Improved Maize Varieties and Inorganic Fertilizer in Kenya. poverty traps: evidence from Ethiopia. J. Dev. Econ. 96–2, 159–173. Environment for Development Discussion Paper Series 14–02. Devarajan, S., 2013. Africa’s statistical tragedy. Rev. Income Wealth 59–S1, S9–S15. O’Sullivan, M., Rao, A., Banerjee, R., Gulati, K., Vinez, M., 2014. Leveling the Field: Devereux, S., Sabates-Wheeler, R., Longhurst, R., 2011. Seasonality, Rural Improving Opportunities for Women Farmers in Africa. World Bank Group, Livelihoods and Development. Routledge, London. Washington D.C. Dillon, B., Barrett, C.B., 2017. Agricultural factor markets in Sub-Saharan Africa: an Palacios-Lopez, A., Christiaensen, L., Galindo, C., 2017. Are the Majority of Rural updated view with formal tests for market failure. Food Policy 67, 64–77. Households Net Food Buyers? World Bank: mimeographed. Doss, C., Raney, T., Anriquez, G., Croppenstedt, A., Gerosa, S., Lowder, S., Matuscke, I., Palacios-Lopez, A., Christiaensen, L., Kilic, T., 2017b. How much of the labor in Skoet, J., 2011, The role of women in agriculture. FAO-ESA Working Paper No. African agriculture is provided by women? Food Policy 67, 52–63. 11-02. Retrieved from . Demographic Research Monographs. Springer, Berlin. Geertz, C., 1963. Agricultural Involution. University of California Press, Berkeley, Los Ravallion, M., 1990. Rural welfare effects of food price changes under induced wage Angeles and London. responses: theory and evidence for Bangladesh. Oxford Economic Papers 42, Gilbert, C.L., Christiaensen, L., Kaminski, J., 2017. Food price seasonality in Africa: 574–585. measurement and extent. Food Policy 67, 119–132. Rodrik, D., 2016. Premature deindustrialization. J. Econ. Growth 21, 1–33. Gollin, D., Jedwab, R., Vollrath, D., 2016. Urbanization with and without Ruthenberg, H.P., 1980. Farming Systems in the Tropics. Clarendon, Oxford. Industrialization. J. Econ. Growth 21, 35–70. Sheahan, M., Black, R., Jayne, T.S., 2013. Are Kenyan farmers under-utilizing Gollin, D., Lagakos, D., Waugh, M.E., 2014. The agricultural productivity gap. Quart. fertilizer? Implications for input intensification strategies and research. Food J. Econ. 129–2, 939–993. Policy 41, 39–52. Ilukor, J. Kilic, T., Stevenson, J., Gourlay, S., Kosmowski, F., Kilian, A., Serumaga, J., Sheahan, M., Barrett, C.B., Goldvale, C., 2016. Human Health and Pesticide Use in Asea, G., 2017, Do we have an ID? Crop variety identification and seed quality Sub-Saharan Africa, mimeographed. assessment in economic research, mimeographed. Sheahan, M., Barrett, C.B., 2017. Ten striking facts about agricultural input use in Ingelaere, B., Christiaensen, L., De Weerdt, J., Kanbur, R., 2017. Urbanisation, Sub-Saharan Africa. Food Policy 67, 12–25. Secondary Towns and the Expanding Horizon of Rural-urban Migrants in Snapp, S., Jayne, T.S., Mhango, W., Benson, T., Ricker-Gilbert, J., 2014. Maize- Tanzania, mimeographed. nitrogen response in Malawi’s smallholder production systems. Malawi, Ivanic, M., Martin, W., 2008. Implications of higher global food prices for poverty in Strategy support program working paper. low-income countries. Agric. Econ. 39, 405–416. Stephens, E.C., Barrett, C.B., 2011. Incomplete credit markets and commodity Jacoby, H.G., 2016. Food prices, wages, and welfare in rural India. Econ. Inq. 54–1, marketing behavior. J. Agric. Econ. 62–1, 1–24. 159–176. Tamru, S., Minten, B., Alemu, D., Bachewe, F., 2016. The Rapid Expansion of Jayne, T.S., Mather, D., Mason, N.M., Ricker-Gilbert, J., 2013. How do fertilizer Herbicide Use in Smallholder Agriculture in Ethiopia: Patterns, Drivers, and subsidy programs affect total fertilizer use in Sub-Saharan Africa? Crowding Implications. Ethiopia Strategy Support Program Working Paper 94. Ethiopian out, diversion, and benefit/cost assessments. Agric. Econ. 44–6, 687–703. Development Research Institute and International Food Policy Research Jayne, T.S. et al., 2016. Africa’s changing farm size distribution patterns: the rise of Institute: Addis Ababa, Ethiopia. medium-scale farms. Agric. Econ. 47 (supplement), 197–214. UNESCO (United Nations Educational, Scientific and Cultural Organization), 2015, Jerven, M., 2013. Poor Numbers: How We Are Misled by African Development Education for All 2000–2015: Achievements and Challenges. EFA Global Statistics and What to Do About It? Cornell University Press, Ithaca, New Monitoring Report 2015: Paris. York. Udry, C., 1999. Efficiency in Market Structure: Testing for Profit Maximization in Jerven, M., 2016. Research note: Africa by numbers: reviewing the database African Agriculture. In: Ranis, G., Raut, L.K. (Eds.), Trade, Growth and approach to studying African economies. African Affairs 115–459, 342–358. Development: Essays in Honor of T.N. Srinivasan, Elsevier: Amsterdam. Kaminski, J., Christiaensen, L., Gilbert, C.L., 2016. Seasonality in local food markets World Bank Group, 2017. World Development Indicators 757. Consulted 20 January 2017. World Bank Group: Washington D.C. LaFave, D., & Thomas, D., 2014. Farms, families, and markets: new evidence on Yilma, T., Berger, T., 2006. Complementarity between Irrigation and Fertilizer agricultural labor markets. National Bureau of Economic Research Paper 20699. Technologies – A Justification for Increased Irrigation Investment in the Less- Liverpool-Tasie, L.S.O., 2016. Is fertilizer use inconsistent with expected profit Favored Areas of SSA, Contributed paper for presentation at the International maximization in sub-Saharan Africa? The case of rice in Nigeria. J. Agric. Econ. Association of Agricultural Economists Conference, Gold Coast, Australia, 12–18 http://dx.doi.org/10.1111/1477-9552.12162. August, 2006.